dalance/amber | ProgramBench

A code search / replace tool

565

Generated Behavioral Tests

100.0%

Best Score

Claude Opus 5.5 (xhigh)

Results by Model

Hover for details · Click a model to open it · Drag to zoom the x-axis · Shift-drag to pan · The line marks the Pareto frontier (best score per cost)

30 runs

#	Model	Score	Cost	Calls
1	Claude Opus 5.5 (xhigh)	100.0%	$29.35	370	trace →
2	Muse Wonder Full	100.0%	TBA	395	trace →
3	Muse Wonder (xhigh)	99.5%	$27.74	336	trace →
4	GPT 6 Astra (max)	96.5%	$9.99	57	trace →
5	Claude Opus 5 (xhigh)	94.9%	$43.90	280	trace →
6	Claude Fable 5.1 (xhigh)	94.3%	$41.99	226	trace →
7	Muse Spark 1.3 (max)	91.9%	$3.33	232	trace →
8	Claude Opus 4.8 (xhigh)	89.4%	$28.09	164	trace →
9	Muse Spark 1.3 (xhigh)	87.3%	$1.73	282	trace →
10	GPT-5.6 Sol (xhigh)	86.9%	$4.37	47	trace →
11	Claude Opus 4.7 (xhigh)	74.3%	$16.71	329	trace →
12	GPT 5.5	72.6%	$0.78	20	trace →
13	GPT-5.6 Sol	71.9%	$0.88	18	trace →
14	Gemini 3.7 Flash	71.9%	$0.72	91	trace →
15	Gemini 3.5 Flash	71.2%	$2.36	116	trace →
16	Muse Spark 1.2 (xhigh)	71.0%	$0.73	125	trace →
17	Claude Sonnet 4.6	71.0%	$23.27	507	trace →
18	Muse Spark 1.1 (xhigh)	70.1%	$0.39	121	trace →
19	GLM-5.2	69.6%	$16.12	162	trace →
20	Claude Opus 4.6	68.7%	$14.47	329	trace →
21	Claude Opus 4.7	67.8%	$2.97	96	trace →
22	GPT 5.4	67.8%	$4.75	1,000	trace →
23	Gemini 3 Flash	64.1%	$0.30	126	trace →
24	Gemini 3.6 Flash	54.3%	$3.60	154	trace →
25	Gemini 3.1 Pro	52.9%	$0.93	91	trace →
26	Claude Haiku 4.5	46.4%	$0.33	81	trace →
27	GPT 5.5 (xhigh)	23.2%	$4.70	77	trace →
28	GPT 5.5 (high)	22.3%	$2.07	44	trace →
29	GPT 5 mini	12.4%	$0.01	8	trace →
30	GPT 5.4 mini	3.4%	$0.02	7	trace →

Click a row to replay how that model rebuilt this program, or the model name to open its full run